Re: Layer Shift Detection Not Working
By pinching the rails you mean you had your finger on the smooth rod?
Or are you pinching the X-axis belt?
I'm asking because having your finger on the smooth rod and letting the extruder hit it seems to be the best way to test it.
I'm pinching the smooth rod and letting the carriage hit my finger.
I love how there's a right and a wrong way to test if the skipped step detection is working... "No, you cannot PUSH the carriage, you have to let it hit a stationary target" 🙂
Re: Layer Shift Detection Not Working
By pinching the rails you mean you had your finger on the smooth rod?
Or are you pinching the X-axis belt?
I'm asking because having your finger on the smooth rod and letting the extruder hit it seems to be the best way to test it.
I'm pinching the smooth rod and letting the carriage hit my finger.
I love how there's a right and a wrong way to test if the skipped step detection is working... "No, you cannot PUSH the carriage, you have to let it hit a stationary target" 🙂
It makes sense.... the printer is smart.... but not that smart xD... It detects when the nozzle crashes into something (shifting layers because of that)...
Pushing the x-carriage to the side is nothing like crashing the nozzle into something! Thats just a move the X-axis made that the printer doesnt even acknowledge as such.
Re: Layer Shift Detection Not Working
Why would you treat any move by any axis that wasn't intentional as anything but an error? It shouldn't matter if it was a push vs a stationary object; it sounds like the detection code isn't as robust as it ideally should be yet. I'll have to go check out GitHub, I've never really looked at the firmware on these things. I'm assuming it's looking for skipped steps, so either path blocks or pushes in the opposite direction of intended movement should work, as long as the carriage is in motion (has a called move).
Re: Layer Shift Detection Not Working
Why would you treat any move by any axis that wasn't intentional as anything but an error? It shouldn't matter if it was a push vs a stationary object; it sounds like the detection code isn't as robust as it ideally should be yet. I'll have to go check out GitHub, I've never really looked at the firmware on these things. I'm assuming it's looking for skipped steps, so either path blocks or pushes in the opposite direction of intended movement should work, as long as the carriage is in motion (has a called move).
That is valid and it would be ideal to work like that.
What i meant to say is that it makes sense that you need to test it they way they say because thats the way they design the detection.
Re: Layer Shift Detection Not Working
It's not even the design- it's the nature of how it works. Shift detection is likely looking for abnormally high currents encountered when the stepper has to push harder than normal (like when the head hits something). The stepper driver cannot detect something like being pushed in a direction it is ALREADY going as the current doesn't increase in this case- hence no way to properly detect it.
Re: Layer Shift Detection Not Working
Another print ruined by layer shifts... This time twice in one print, and in Y-axis. No layer shift detection. Latest firmware (b143), Slic3r 1.38.6 "mk3 otpimal" profile, standard/non-stealth settings. One shift is about 1mm, the other 3mm. Belts thight (but not too thight), everything runs smoothly...
I have around 12 days of total print time on my mk3, and I've had about 4-5 problems with unexplained and undetected layer shifts, across three different firmwares, at least three different materials, both Cura 3.1.0 and several different versions of Slic3r... And both X and Y axis. Any ideas?
Re: Layer Shift Detection Not Working
I think you guys need to disable the silent printing mode
The trinamic drivers really have two modes.. one that does the super silent method but that is designed for things like opening a vent or something that does not require accuracy. 3d printers should never use this mode
The other mode is 16 microstep 256 interpolation which is the normal mode suitable for 3d printers. See the TMC2100 datasheet.
TLDR if the printer has motor/driver/noise options choose the noisiest setting.
Re: Layer Shift Detection Not Working
I think you guys need to disable the silent printing mode
The trinamic drivers really have two modes.. one that does the super silent method but that is designed for things like opening a vent or something that does not require accuracy. 3d printers should never use this mode
The other mode is 16 microstep 256 interpolation which is the normal mode suitable for 3d printers. See the TMC2100 datasheet.
TLDR if the printer has motor/driver/noise options choose the noisiest setting.
All my problems has been with the silent/stealth mode disabled. The documentation is quite clear on the fact that layer shift detection does not work in silent/stealth mode.
Re: Layer Shift Detection Not Working
I think you guys need to disable the silent printing mode
The trinamic drivers really have two modes.. one that does the super silent method but that is designed for things like opening a vent or something that does not require accuracy. 3d printers should never use this mode
The other mode is 16 microstep 256 interpolation which is the normal mode suitable for 3d printers. See the TMC2100 datasheet.
TLDR if the printer has motor/driver/noise options choose the noisiest setting.
All my problems has been with the silent/stealth mode disabled. The documentation is quite clear on the fact that layer shift detection does not work in silent/stealth mode.
have you seen it happen?
My observation is that there is a difference in both slic3r and cura where it runs at a very low speed for some things and possibly in the slow mode a step or two can be skipped without creating a large current change. I would also think the tighter the belts are it would be hard to tell that there is a shift in resistance/current.
all my layer shifts (19 so far) have tried to recover. most were during failed prints with petg where it was balling up and hardening into an unconquerable obstruction. i've had a couple with abs and of those 1 or 2 were not able to recover and 1 shifted all of those were sliced in cura.
I have noticed that prusacontrol does the same thing as the sample prints and z hops all the time which seems to help with not knocking over the support when travelling over it.
Re: Layer Shift Detection Not Working
I have 2 MK3s. We've been printing on them for 2 weeks now. We've gone through about 6 spools now, with about half of the prints failing from layer shift issues. The layer shifting is usually in the y axis, but sometimes, like in these photos, it's in both axes. This one is really weird, with the failure in both axes, at the same time.
Normal mode, not stealth
Slic3r
I had a similar issue with a Tevo Little Monster printer and fixed it by reducing the motor drive current. This seems like a motor drive current problem to me, maybe needing a higher or lower current to fix it. The only problem with that is I don't think it would be able to maintain the 200mm/s as advertised.
Re: Layer Shift Detection Not Working
Had my first layer shift today along the x-axis. Using the latest firmware on normal mode. Just tried tightening my belt a bit more, will see if that helps.
Re: Layer Shift Detection Not Working
Another user with layer shift problems here.
With it enabled in normal mode its ruining my prints. Printer is reporting 15 errors although ive seen nothing to indicate there is any need for the layer shift to kick in. Its like its to sensitive or something, is there any way to adjust that?
Just to clarify with layer shift enabled its actually causing a massive layer shift in the print, with it disabled the print works perfect.
Ive turned this feature off for now and prints are running great again.
Re: Layer Shift Detection Not Working
I have the same issue. Running the latest firmware (b150). I checked to make sure the steppers were in normal mode and crash detection was set to on. Then after a couple layers I tried obstructing the print by pinching the x rod with fingers. Stepper skipped multiple times but no recovery, it just kept printing. Tried the same on the Y axis with similar results. I was not pushing the axis, just allowing the axis to crash into my fingers. The printer homes just fine and no issues during setup/calibration wizzard. I'm assuming a firmware issue, but I'm open to suggestions if there is anything else I can test.
Bryan
Re: Layer Shift Detection Not Working
I also have detected problems.
I will try to test without shift detection, everything works fine if i try to pinch the bar for the x-axis and it detects the skip and resets, but during printing it does skip at times and at different heights so it does not hit something. My belts are quite tight but not too tight.
Firmware 3.1.1 RC5-150 and prusa slic3r 1.39.1 alpha mk3 profile stealth mode off and all.
Re: Layer Shift Detection Not Working
This one is really weird, with the failure in both axes, at the same time
Same here, layer shift in both axis. I did spot it earlier in the print claiming crash detected, it continued fine but obviously later some other issue. I do not believe it hit anything, false positive. I am seeing layer shift probably every other or every third print. Very frustrating. Going to try with it off but clearly some issue going on somewhere..
Re: Layer Shift Detection Not Working
My overall failure rate is the same as yours.
After two consecutive failures, I started with the detection off.
After 5 hours out of 11 total, everything is running perfectly. I can't say this too loud or the machine will hear me. 😉
It'll take several 10+ hour prints without any shifts to convince me that we found the problem. My fingers are crossed.
Re: Layer Shift Detection Not Working
I have had several small layer shifts in the past month as well, will turn of shift detection on mine and see if that fixes it. hope they let us know if there is a firmware fix for this.
Re: Layer Shift Detection Not Working
Just a quick thought on layer shifting issues. Whilst I await my MK3 to be delivered, I do have some experience with the TMC2208 drivers as a retrofit onto a K8400. https://forum.vellemanprojects.eu/t/tmc2100-drivers-on-k8400/14610/8
One of the things I found they tended to do was to skip steps if:
• Any of the hardware was not perfectly aligned and smooth like butter to move with the belts off.
• Drivers pushing too high a current to the motors
• Not enough cooling of the driver chips
I found them to be so fussy, I've put them on the back burner and saving them for a rebuilt K8400 project in the future.
Now it is quite likely there is still a weird bug in the Marlin firmware where crash detection is causing skipped steps. It is worth just bearing in mind these other three causes however, rather than just blaming bugs in the code.
The running current - on the MK3 that is set in software with the Einsy board, so we cannot mess with that too easily. What we can look at is hardware alignment and cooling. What I have seen from the build instructions - there is a right rats nest of wires in the box housing the Einsy PCB. Whilst the Trinamic drivers sink most of their waste heat into the PCB, if there is poor ventilation over the top of any of the Trinamic chips where cable management is poor, this might just be "the straw that broke the camel's back" and cause the driver to momentarily reset on thermal overload. Something so simple to check, simply by moving some wires around for optimum ventilation.
When I do get my MK3, I'll be getting the Flir thermal imaging camera out to see how hot the drivers are getting - more to satisfy my own curiosity 😀
Re: Layer Shift Detection Not Working
Is it not that these are cases of 'false positive layer shift detection', causing these 'corrections' in your layers, that people are calling 'layer shifts'?
That would imply that the geometry of your printer is causing difficulties to the stepper that it is interpreting as layer shifts?
It seems that as the thread was created about actual crash detection testing, these two subjects have combined and confused things.
Re: Layer Shift Detection Not Working
Ok, so I've tried a little detective work here, I printed a benchy and it got skipped steps around 15,9 millimiters z-height, and after some layers it fixed itself. So i tried printing the exact same gcode but with crash detection turned off and the print came out almost perfect and with no layer shifts.
This got me wondering, so I printed yet another one of the darned benchy's and with crash detection turned on again, and the shifted layers reappears at almost the exact same height as the first one, and corrects itself almost at the exact same height as the first one, so it does seem there is some problems with the crash detection for some reason, when i try to pinch the rails it works just fine though.
Firmware 3.1.1 RC5-150 and prusa slic3r 1.39.1 alpha mk3 profile stealth mode off and all.
(sorry if pictures does not really work, I'm not that good at adding things in posts on forums yet)